4b4c0fe86c
2005-07-26 Federico Mena Quintero <federico@ximian.com> * perf/: New directory with the start of a framework for testing performance in GTK+. * Makefile.am (SRC_SUBDIRS): Added the perf directory. * configure.in (AC_OUTPUT): Generate perf/Makefile.
128 lines
3.6 KiB
Plaintext
128 lines
3.6 KiB
Plaintext
README for gtk+/perf
|
|
--------------------
|
|
|
|
This is a framework for testing performance in GTK+. For GTK+, being
|
|
performant does not only mean "paint widgets fast". It also means
|
|
things like the time needed to set up widgets, to map and draw a
|
|
window for the first time, and emitting/propagating signals.
|
|
|
|
The following is accurate as of 2005/07/26.
|
|
|
|
|
|
Using the framework
|
|
-------------------
|
|
|
|
Right now the framework is very simple; it just has utility functions
|
|
to time widget creation, drawing, and destruction. To run such a
|
|
test, you use the functions in timers.h. You can call this:
|
|
|
|
timer_time_widget (create_func, report_func, user_data);
|
|
|
|
You must provide the create_funcn and report_func callbacks.
|
|
|
|
The create_func:
|
|
|
|
This simply creates a toplevel window with some widgets inside it.
|
|
It is important that you do not show any of the widgets; the
|
|
framework will call gtk_widget_show_all() on the toplevel window
|
|
automatically at the right time.
|
|
|
|
The report_func:
|
|
|
|
This function will get called when timer_time_widget() reaches an
|
|
interesting point in the lifecycle of your widget. See timers.h and
|
|
the TimerReport enumeration; this is what gets passed as the
|
|
"report" argument to your report_func. Right now, your function
|
|
will be called three times for each call to timer_time_widget():
|
|
|
|
1. With report = TIMER_REPORT_WIDGET_CREATION. A timer gets
|
|
started right before timer_time_widget() calls create_func,
|
|
and it gets stopped when your create_func returns. This
|
|
measures the time it takes to set up a toplevel window (but
|
|
not show it).
|
|
|
|
2. With report = TIMER_REPORT_WIDGET_SHOW. A timer gets started
|
|
right before timer_time_widget() calls gtk_widget_show_all()
|
|
on your toplevel window, and it gets stopped when the window
|
|
has been fully shown and painted to the screen.
|
|
|
|
3. With report = TIMER_REPORT_WIDGET_DESTRUCTION. A timer gets
|
|
started right before timer_time_widget() calls
|
|
gtk_widget_destroy() on your toplevel window, and it gets
|
|
stopped when gtk_widget_destroy() returns.
|
|
|
|
As a very basic example of using timer_time_widget(), you can use
|
|
something like this:
|
|
|
|
----------------------------------------------------------------------
|
|
#include <stdio.h>
|
|
#include <gtk/gtk.h>
|
|
#include "timers.h"
|
|
|
|
#define ITERS 20
|
|
|
|
static GtkWidget *
|
|
create_cb (gpointer data)
|
|
{
|
|
GtkWidget *window;
|
|
|
|
window = gtk_window_new (GTK_WINDOW_TOPLEVEL);
|
|
/* ... fill the window with widgets, and don't show them ... */
|
|
|
|
return window;
|
|
}
|
|
|
|
static void
|
|
report_cb (TimerReport report, gdouble elapsed, gpointer data)
|
|
{
|
|
const char *type;
|
|
|
|
switch (report) {
|
|
case TIMER_REPORT_WIDGET_CREATION:
|
|
type = "widget creation";
|
|
break;
|
|
|
|
case TIMER_REPORT_WIDGET_SHOW:
|
|
type = "widget show";
|
|
break;
|
|
|
|
case TIMER_REPORT_WIDGET_DESTRUCTION:
|
|
type = "widget destruction";
|
|
break;
|
|
}
|
|
|
|
fprintf (stderr, "%s: %g sec\n", type, elapsed);
|
|
}
|
|
|
|
int
|
|
main (int argc, char **argv)
|
|
{
|
|
int i;
|
|
|
|
gtk_init (&argc, &argv);
|
|
|
|
for (i = 0; i < ITERS; i++)
|
|
timer_time_widget (create_cb, report_cb, NULL);
|
|
|
|
return 0;
|
|
}
|
|
----------------------------------------------------------------------
|
|
|
|
|
|
Getting meaningful results
|
|
--------------------------
|
|
|
|
Getting times for widget creation/drawing/destruction is interesting,
|
|
but how do you actually find the places that need optimizing?
|
|
|
|
Why, you run the tests under a profiler, of course.
|
|
|
|
FIXME: document how to do this.
|
|
|
|
|
|
Feedback
|
|
--------
|
|
|
|
Please mail your feedback to Federico Mena-Quintero <federico@novell.com>.
|
|
This performance framework is a work in progress.
|